Objectives To report the clinical features and prognosis of drug-associatedacute respiratory distress syndrome (ARDS). Design A retrospective analysis of data collected during a prospective cohort study. Setting Intensive care unit in a teaching hospital. Participants A total of 197 Japanese patients with ARDS diagnosed by the Berlin definition who were admitted to the Division of Respiratory Medicine from October 2004 to December 2015 were enrolled in the study and were classified as two groups according to their causes: a drug-associated ARDS group (n=27) and a non-drug-associated ARDS group (n=170). Primary outcome measure is 28-day mortality, and the secondaryoutcome measure is ventilator-free days. Results The Acute Physiology and Chronic Health Evaluation II scores were significantly lower in the drug-associated ARDS group than in the non-drug-associated ARDS group (median (IQR): 18.0 (16.5–21.0) vs 23.0 (18.0–26.0), p<0.001), and the arterial oxygen tension/fractional inspired oxygen ratio was higher (148.0 (114.1–177.5) vs 101.0 (71.5–134.0), p=0.003). In the drug-associated ARDS group, although high-resolution CT scores indicative of the extent of fibroproliferation (301.6 (244.1–339.8) vs 208.3 (183.4–271.6), p<0.001), serum lactate dehydrogenase levels (477 (365–585) vs 322 (246–434), p=0.003) and the McCabe scores (score 1/2/3, n (%): 20 (74)/4 (15)/3 (11)vs154 (91)/7 (4)/9 (5), p=0.04) were significantly higher, ventilator weaning was earlier (p<0.001) and 28-day mortality was better (p=0.043). After adjusting for potentially confounding covariates, drug-associated ARDS group was associated with lower 28-day mortality (adjusted HR (HR) 0.275; 95% CI 0.106 to 0.711; p=0.008). Conclusions Although more severe lung damage with fibroproliferation was observed in patients with drug-associated ARDS, ventilator weaning was earlier, and their prognosis was better than the others. Further well-designed prospective studies are needed.